DP-900 domain - 29% of the exam

Describe an Analytics Workload on Azure

Describe an Analytics Workload on Azure is 29% of the DP-900 exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

76
Practice questions
3
Objectives
29
Easy
47
Medium

Where people go wrong

  • Swapping batch and stream processing

    The wrong options exchange the two processing styles. Some give stream processing the complete dataset, others give batch the per-event handling, and others put the rolling time window on batch. Batch processes all the data in a prepared, finite set, with latency of hours, and suits complex analytics. Stream processing acts on each event as it arrives, sees only recent or windowed data, has latency of seconds or milliseconds, and suits simple aggregates such as rolling averages.

    See it in a sample question
  • Matching the wrong Synapse resource model to the workload

    The wrong options pick the model that reverses the stated need. A steady nightly warehouse report is offered serverless, and ad-hoc queries over lake files are offered a dedicated pool or Spark pools. A dedicated SQL pool reserves processing power for data in SQL tables, giving predictable performance and cost. A serverless SQL pool is an always-available on-demand endpoint for unplanned or bursty workloads, and it queries CSV and Parquet files in the lake with nothing reserved.

    See it in a sample question
  • Choosing the wrong ingestion source for an event stream

    The wrong options offer IoT Hub for general application events, Kafka or Data Lake Storage Gen2 for an ordered event queue, and Power BI, OneLake or Azure SQL Database as an ingestion source. Event Hubs queues event data so that each event is processed in order, exactly once. IoT Hub is the variant optimised for Internet-of-Things device telemetry. Kafka is a valid open-source ingestion option most often paired with Spark, Data Lake Storage Gen2 is mainly batch storage, and Power BI, OneLake and SQL Database are sinks for processed results.

    See it in a sample question
  • Swapping Real-Time Intelligence components

    The wrong options hand a job to the neighbouring component: Eventhouse is offered for routing streams, Eventstream for triggering actions, Activator for storing data, and a Real-Time Dashboard for any of these. Eventstream ingests, routes and transforms streaming data. Eventhouse stores time-series and event data that is queried with KQL. Activator fires automated actions when streaming data meets a condition. A Real-Time Dashboard only visualises live data.

    See it in a sample question
  • Placing a Power BI task in the wrong component

    The wrong options give authoring to the service, the phone app or a dashboard, and give refresh scheduling and sharing to Desktop or Model view. Power BI Desktop is the Windows application for importing data, building the model and creating reports, and Model view holds relationships and hierarchies. The Power BI service is the cloud location where reports are published, refreshes are scheduled and content is shared. The phone app only consumes published content.

    See it in a sample question

Objectives in this domain

Sample questions from this domain

Free sampleDescribe an Analytics Workload on Azuremedium

Compared with batch processing, what is typical of the latency seen in stream processing?

  • AStream latency is usually several hours, while batch latency is only seconds.
  • BBoth stream and batch share the same latency once the source queue is filled.
  • CStream latency grows with dataset size, while batch latency stays constant.
  • DStream latency is in the order of seconds or milliseconds, while batch is hours. Correct
Batch latency is measured in hours; stream latency is in seconds or milliseconds. Latency is the time taken for data to be received and processed; for batch it is typically a few hours, while stream processing occurs immediately with latency in the order of seconds or milliseconds.

Why A is wrong: This reverses the two: batch latency is typically a few hours, not stream latency.

Why B is wrong: The two differ markedly in latency, the whole point of choosing stream processing for time-critical work.

Why C is wrong: Latency for stream processing is consistently low, and batch latency is the higher of the two, not the steadier one.

Why D is correct: Correct. Latency is the time taken for data to be received and processed; for batch it is typically a few hours, while stream processing occurs immediately with latency in the order of seconds or milliseconds.

Free sampleDescribe an Analytics Workload on Azuremedium

A retail analytics team runs the same nightly reporting workload against a structured data warehouse every day, and wants predictable performance and cost with processing power reserved in advance. Which Azure Synapse Analytics resource model fits this requirement?

  • AA dedicated SQL pool, which reserves processing power for data stored in SQL tables to give predictable performance and cost Correct
  • BA serverless SQL pool, which uses the always-available on-demand endpoint to handle unplanned or bursty query workloads
  • CAn Apache Spark pool, which provides autoscaling clusters for big-data preparation and machine learning code
  • DA Data Explorer pool, which is optimized for interactive log and time-series analytics over system-generated logs
Dedicated SQL pools reserve capacity for predictable, planned warehouse workloads. For predictable performance and cost, Synapse SQL lets you create dedicated SQL pools that reserve processing power for data stored in SQL tables.

Why A is correct: Correct. For predictable performance and cost, Synapse SQL lets you create dedicated SQL pools that reserve processing power for data stored in SQL tables.

Why B is wrong: Serverless is for unplanned or bursty workloads, not a steady nightly job where reserved, predictable capacity is wanted.

Why C is wrong: Spark pools target big-data prep, engineering, and ML, not predictable T-SQL warehouse reporting on reserved capacity.

Why D is wrong: Data Explorer is for log and time-series analytics, not a structured nightly SQL warehouse workload.

Free sampleDescribe an Analytics Workload on Azuremedium

Which of the following are services commonly used to ingest data as a source for stream processing on Azure? (Select 3 answers)

  • AAzure Event Hubs, managing queues of event data so each event is processed in order and exactly once. Correct
  • BAzure IoT Hub, an ingestion service optimized to manage event data from Internet-of-Things devices. Correct
  • CApache Kafka, an open-source data ingestion solution commonly used together with Apache Spark. Correct
  • DMicrosoft Power BI, a tool used to generate real-time data visualizations in reports and dashboards.
  • EAzure SQL Database, used to persist processed stream results in a table for querying and analysis.
Stream-ingestion sources include Event Hubs, IoT Hub, and Kafka; Power BI and SQL Database sit at the output (sink) side. Event Hubs, IoT Hub, and Apache Kafka (along with ADLS Gen2) are the services the grounding lists for ingesting data for stream processing; Power BI and SQL Database are sinks.

Why A is correct: Correct. Azure Event Hubs, managing queues of event data so each event is processed in order and exactly once is one of the keyed answers. Event Hubs, IoT Hub, and Apache Kafka (along with ADLS Gen2) are the services the grounding lists for ingesting data for stream processing; Power BI and SQL Database are sinks.

Why B is correct: Correct. Azure IoT Hub, an ingestion service optimized to manage event data from Internet-of-Things devices is one of the keyed answers. Event Hubs, IoT Hub, and Apache Kafka (along with ADLS Gen2) are the services the grounding lists for ingesting data for stream processing; Power BI and SQL Database are sinks.

Why C is correct: Correct. Apache Kafka, an open-source data ingestion solution commonly used together with Apache Spark is one of the keyed answers. Event Hubs, IoT Hub, and Apache Kafka (along with ADLS Gen2) are the services the grounding lists for ingesting data for stream processing; Power BI and SQL Database are sinks.

Why D is wrong: Power BI is listed as a sink for visualization, not as an ingestion source.

Why E is wrong: Azure SQL Database is listed as a sink for processed results, not as an ingestion source.

Free sampleDescribe an Analytics Workload on Azuremedium

Inside Microsoft Fabric Real-Time Intelligence, an organisation wants to automatically trigger an action whenever incoming streaming data meets a defined condition, such as a temperature crossing a threshold. Which component provides this capability?

  • AEventhouse, the database optimised for time-series and event data queries
  • BEventstream, used to ingest, route, and transform incoming streaming data
  • CReal-Time Dashboard, which presents live visualisations of streaming data
  • DActivator, which fires automated actions when streaming conditions are met Correct
Activator is the Fabric Real-Time Intelligence component that triggers automated actions on defined streaming conditions. Activator is described as the Real-Time Intelligence component that triggers automated actions when streaming data meets defined conditions.

Why A is wrong: Eventhouse stores and queries event data with KQL but does not itself trigger automated actions.

Why B is wrong: Eventstream handles ingestion, routing, and transformation, not condition-based action triggering.

Why C is wrong: A Real-Time Dashboard visualises live data for people to view, rather than firing automated reactions.

Why D is correct: Correct. Activator is described as the Real-Time Intelligence component that triggers automated actions when streaming data meets defined conditions.

Free sampleDescribe an Analytics Workload on Azureeasy

After building reports in Power BI Desktop, a team needs a cloud location where business users can interact with the reports, schedule data refreshes, and share content. Which Power BI component provides this?

  • APower BI Desktop, the Windows tool used to build the original reports
  • BThe Power BI phone app, an application for viewing content on a phone
  • CA semantic model, the structure that defines measures and relationships
  • DThe Power BI service, a cloud service for publishing and sharing reports Correct
The Power BI service is the cloud home for publishing, scheduling refreshes, and sharing. The Power BI service is a cloud service where reports are published and interacted with by business users, where refreshes are scheduled, and where reports are shared.

Why A is wrong: Desktop is for authoring on one machine; it does not host shared cloud reports or scheduled refreshes.

Why B is wrong: The phone app is a consumption client; it does not publish reports or schedule refreshes.

Why C is wrong: A semantic model is the data structure reports use, not a place to publish and share them.

Why D is correct: Correct. The Power BI service is a cloud service where reports are published and interacted with by business users, where refreshes are scheduled, and where reports are shared.

Other domains in this exam

See also the DP-900 cert hub, the study guide, and the cheat sheet.

Examworthy is not affiliated with or endorsed by Microsoft. Original, blueprint-aligned practice material only.